The short answer
MarketPrior gives AI agents and internal tools structured, cited company data through an MCP server at /mcp (Streamable HTTP, Bearer auth) and a REST API under /v1 with an OpenAPI spec. Agents can search companies, start diligence runs, read report sections with their evidence and freshness, query the lead trigger feed, work with contacts and exports, and read alerts — all with scoped API keys and the same paywall, quotas and credits as the app.
At a glance
- MCP
- Server at /mcp — Streamable HTTP, Bearer auth
- REST
- /v1 with an OpenAPI spec at /v1/openapi.json
- Keys
- Scoped: read, run diligence, monitor, leads, reveal, export
- Key storage
- Shown once; only a SHA-256 hash is stored
- Responses
- Sections with evidence, citations and freshness
- Limits
- Shared rate limits and credits across REST and MCP
Why do agents need cited data?
Language models are fluent but not reliable about specific companies: names collide, facts go stale, and a confident wrong answer is worse than none. Grounding an agent in a tool that returns sourced, dated findings lets it answer with citations — and lets a person check the answer.
What can an agent do with MarketPrior?
- Resolve and search companies, then read a full report or a single section with its evidence.
- Start a diligence run as a job and poll it until it finishes.
- Read the lead trigger feed and run ICP searches with the same filters as the app.
- Work with masked contacts, reveals and exports within the key’s scopes and credits.
- Read monitoring alerts (read-only over MCP; list and acknowledge over REST).
How do you connect an agent?
- 1
Create a scoped API key
In Settings → API keys, with only the scopes the agent needs.
- 2
Add the MCP server to your client
Point any Streamable-HTTP MCP client at /mcp with the key as a Bearer token.
- 3
Or call the REST API
Use the OpenAPI spec to generate a client, or call /v1 directly.
- 4
Respect the paywall and limits
Locked sections return 402 report_locked; rate limits are shared across REST and MCP.